At about midday on the same day, the pilot found that a revised TAF now only required the carriage of an additional 60 minutes of holding fuel for arrival at Norfolk Island. The forecast, issued at 1133, indicated that from 1200 to 2300 there would be broken (more than 4/8ths) cloud at 2,500 ft, scattered (4/8ths or less) cloud at 500 ft, and visibility reduced to 8,000 m in rain showers. In addition, there would be periods of up to 60 minutes (TEMPO) of broken cloud at 400 ft, with visibility reduced to 4,000 m in drizzle. The minima for a straight-in landing approach to runway 11 required not more than 4/8ths cloud below 750 ft, with a visibility of not less than 2,700 m. The alternate minima required not more than 4/8ths cloud below 1,069 ft, and visibility not less than 4,400 m.

The aircraft subsequently departed Lord Howe Island carrying the additional holding fuel. Before reaching his calculated point of no return (PNR) at about 1545 the pilot received two meteorological reports (METARs) of conditions at Norfolk Island. Both reports described conditions below the published alternate and landing minima. Moreover, an amended TAF for Norfolk Island issued at 1420 indicated that from 1400 to 2300 the amount of cloud at 500 ft had increased to broken (more than 4/8ths). Based on that amendment the pilot was required to nominate an alternate aerodrome for the flight. However, he did not request or receive the amended weather forecast before passing the PNR, as required by the company operations manual.

As the flight approached Norfolk Island, the pilot was advised of reducing visibility. He conducted an instrument approach to runway 11 but did not become visual, and so carried out a missed approach. In conditions of deteriorating visibility the pilot then flew several approaches to runway 04 by descending over the sea until he was visual, at a height of approximately 500 ft, before tracking inbound on the instrument approach VOR (a navigation aid) radial. On the first two approaches, the pilot was unable to see the aerodrome and conducted a missed approach.

For the third approach, the pilot requested the aerodrome manager to position himself on the runway threshold in order to indicate by radio the position of the aircraft in relation to the runway centreline. The manager did this and advised the pilot that the aircraft had passed to the right of the centreline. On the fourth approach, the pilot sighted the precision approach path indicator lights and landed on the runway. Visibility at the time of landing was recorded as 800 m in fog.

The pilot later reported that he had made the decision to continue beyond the PNR, based on the forecast obtained at the time he had completed his flight planning. He believed that it was still the current forecast as he had not been advised during the flight of any amendments to that forecast.

Several factors were considered significant to the development of this occurrence. The meteorological conditions which developed at Norfolk Island later in the day, were considerably worse than those forecast by the TAFs issued at 1133 and 1420. Both forecasts indicated that conditions would be below the landing minima for periods of up to 60 minutes. In reality, the conditions at Norfolk Island were below the landing minima continuously from 1130 until after 2330. Had the pilot requested the latest TAF before passing the PNR, he would have been obliged to return to Lord Howe Island, the nearest alternate. In addition, the pilot did not use or did not acknowledge the significance of the METAR information he had received in flight to challenge the accuracy of the TAF that he used for his decision making.

As the regulations did not require RPT aircraft conducting flights to Norfolk Island to carry sufficient fuel for flight to an alternate aerodrome, the pilot carried only an additional 60 minutes of fuel in accordance with the operational requirements of the 1133 TAF. Having passed his PNR, the aircraft did not have sufficient fuel to return to Lord Howe Island, and the pilot was subsequently left with limited options when he encountered the subsequent conditions at Norfolk Island.

The Squirrel helicopter was being used to carry an underslung load of operating fireworks during the Perth Australia Day fireworks display. The helicopter's flight path followed the Swan River, remaining clear of spectators. After the fireworks were ignited, some projectiles from the fireworks appeared to pass through or close to the left side of the helicopter's main rotor disc. The helicopter was not damaged.

The helicopter operator had approached the local District Office of the Civil Aviation Safety Authority (CASA) for an approval to conduct the display. The operator was of the understanding that the fireworks were non-projectile and advised this to the local CASA officers. Although the company fitting the fireworks had advised the event promoter that the fireworks included eight-shot Roman Candles, eight-shot Crosette Candles, Flares and Silver Fountains, the event promoter, the local CASA officers, and the operator expected a cascading type display with nothing ejecting from the helicopter's underslung load.

Because it was the first display of its type in Australia, the local CASA officers sought advice from CASA officers in Canberra. Although the Canberra based officers advised against approving the display, the local CASA officers considered that because the fireworks were non-projectile, the display could be conducted safely if the operator met certain guidelines. The helicopter operator was also required to demonstrate that the rig, on which the fireworks were mounted, could be safely flown as an underslung load. The local officers reported that CASA officers in Canberra advised that because the fireworks were being flown as an underslung load, they were not regarded as dangerous goods.

The rig on which the fireworks were mounted, was a large circular metal frame attached to the helicopter by web-type slings. When flown, the rig demonstrated good flying qualities and the local CASA officers reported that the mechanisms securing the fireworks to the rig appeared sound. However, the assessment flight was done without the fireworks attached or a test firing of the fireworks. A Flying Operations Inspector from the local CASA district office granted conditional approval for the flight. The conditions included requirements to remain a minimum distance of 300 m from the shoreline and that the display was not to be flown over any person or boat.

The helicopter lifted the load from a pad near the Swan River and while it was flying along the river, the fireworks were ignited electrically from a control box operated by a pyro-technician sitting in the cabin. The pilot reported that he was surprised when the first fireworks ignited, that they were red flares that ejected from the rig. However, they operated without incident. The cascading-type fireworks also operated without incident but when the Roman Candles fired, some of the shots fired upwards towards the helicopter. The pilot reported that he had felt the rig moving in response to the igniting fireworks but this movement did not affect the controllability of the helicopter. He was unaware that any of the shots had come close to the helicopter until the copilot later reported that some had appeared to do so.

After the helicopter landed, it was found that eight rounds of the Roman Candles had dislodged during the flight and fallen from the rig. The recoil generated by the shots ejecting from the rounds appeared to have caused the rounds to move up and out of the securing straps. As the rounds fell from the rig, they tumbled and the shots continued, some of which passed close to the helicopter. The company that fitted the fireworks reported that the strength of the recoil generated by the Roman Candle had been underestimated.

During the investigation, it became apparent that there were differing opinions as to whether an underslung load was considered to be part of the helicopter with respect to dangerous goods requirements. CASA subsequently informed the investigation that anything attached to an aircraft is considered to be part of the aircraft and that dangerous goods carried as an underslung load must be treated no differently from dangerous goods carried inside the aircraft. CASA also advised that dangerous goods carried differently to that which is required by the Civil Aviation Regulations must be subject to written permission issued by CASA.

There was a misunderstanding among the pyro-technicians, event organiser, helicopter operator and local CASA officers in relation to the types of firework being carried by, and fired from, the helicopter. As a result, the approval given by CASA for the display was based on incorrect information. The mechanisms that secured the Roman Candle rounds to the rig did not prevent the rounds from dislodging from the airframe and were therefore inadequate for the purpose.

History of the flight

Uzu Air conducted passenger and freight operations between Horn Island and the island communities in the Torres Strait. It operated single-engine Cessna models 206 and 208 aircraft, and twin-engine Britten Norman Islander aircraft.

On the morning of the accident, the pilot flew a company Cessna 206 aircraft from Horn Island to Yam, Coconut, and Badu Islands, and then returned to Horn Island. The total flight time was about 93 minutes.

The pilot's schedule during the afternoon was to fly from Horn Island to Coconut, Yam, York, and Coconut Islands and then back to Horn Island, departing at 1330 eastern standard time. The flight was to be conducted in Islander, VH-XFF. Three passengers and about 130 kg freight were to be carried on the Horn Island - Coconut Island sector. Another company pilot had completed three flights in XFF earlier in the day for a total of 1.9 hours. He reported that the aircraft operated normally.

Witnesses at Horn Island reported that the preparation for the flight, and the subsequent departure of the aircraft at 1350, proceeded normally. The pilot of another company aircraft heard the pilot of XFF report 15 NM SW of Coconut Island at 3,500 ft. A few minutes later, the pilot reported downwind for runway 27 at Coconut Island. Both transmissions sounded normal.

Three members of the Coconut Island community reported that, at about 1410, they were on the beach at the eastern extremity of the island, about 250 m from the runway threshold and close to the extended runway centreline. Their recollections of the progress of the aircraft in the Coconut Island circuit are as follows: the aircraft joined the downwind leg and flew a left circuit for runway 27; the aircraft appeared to fly a normal approach until it passed over their position at an altitude of 200-300 ft; and it then veered left and commenced a shallow climb before suddenly rolling right and descending steeply onto a tidal flat, about 30 m seaward from the high-water mark, and about 200 m from their position.

Injuries to persons

Damage to aircraft

Severe disruption to the outer right wing and nose sections occurred as a result of impact forces. Less significant damage occurred to the outer left wing leading edge. The fuselage fractured just aft of the wing trailing edge. The wing attachment points failed, allowing the wing to rotate forward and partially crush the cockpit/forward cabin area. There was a compression fracture of the upper surface of the left horizontal stabiliser near the inboard end. The outboard end of the left stabiliser had been bent upwards by ground impact.

Other damage

There was no other damage.

Personnel

• Pilot
   

  Age:	27
  Licence category:	Commercial
  Medical certificate:	Class 1 (valid to 27 June 1999)
  Instrument rating:	Command Multi Engine
  Total flying hours:	2,540
  Total on type:	197
  Total last 90 days:	205
  Total last 24 hours:	2
  Last flight check:	12 November 1998

• Flying experience and qualifications
  The pilot began flying in 1990 and gained a Private Pilot (Aeroplane) Licence on 21 March 1991. She was issued with a Commercial Pilot Licence on 18 August 1994 and gained a Command Multi-engine Instrument Rating on 21 October 1996. On 1 February 1995, the pilot qualified as a Grade 3 Fixed Wing flying instructor, and gained a Grade 1 instructor rating on 25 September 1997. She was issued with a multi-engine training approval on 30 March 1998.

  As well as being qualified to fly Islander aircraft, the pilot held endorsements on a number of other twin-engine aircraft, including Aero Commander, Beechcraft Baron, Cessna 310, Piper's Navajo, Seneca, and Seminole.

  The pilot completed her endorsement on the Islander on 10 September 1998 and a proficiency check on 16 September 1998. The endorsement and check reports indicated that the pilot operated the aircraft at a high standard, and was disciplined and thorough with checks and procedures. No significant deficiencies were recorded. The training included asymmetric handling sequences, one of which was a simulated single-engine go-around from final approach with the wing flaps at the take-off position.

  The pilot completed airfield checks at a number of airstrips in the Torres Strait, including Coconut Island, on 12 November 1998.

• Seven-day history
  The following summary of the pilot's flight and duty times was taken from company records:

  Date	Duty hours	Flight time (hours)
  9 January	0630-1800	4.1
  10 January	day off	nil
  11 January	1200-1800	nil
  12 January	0700-1800	4.8
  13 January	reserve	nil
  14 January	0700-1830	7.3
  15 January	day off	nil

  Associates of the pilot reported that she appeared in good health on the morning of the accident.

• Seat cushions
  The pilot was approximately 157 cm tall. The operator reported that the pilot used two foam-rubber cushions (one on the seat and the other against the seat back) to adjust her seating position to enable her to achieve full movement of the cockpit controls. The seat cushions normally used by the pilot were not found. No person was found who could recall the pilot taking the cushions to the aircraft before the flight. However, the cushions were not at the company office where they were usually stored when not in use. Assuming they were on the aircraft, it is likely that they were lost as a result of the post-accident tidal and/or wind action.

Aircraft information

• Significant particulars
   

  Registration:	VH-XFF
  Manufacturer:	Britten Norman Pty Ltd
  Model:	BN2A-26 Islander
  Serial number:	C763
  Country of manufacture:	United Kingdom
  Engines:	Lycoming O-540-E4C5

• Certificate of airworthiness
   

  Number:	CS/34
  Issued:	18 December 1989
  Category of operation:	Normal

• Certificate of registration
   

  Holder:	Uzu Air Pty Ltd
  Number:	CNS/00034/04
  Issued:	6 January 1994

• Maintenance release
   

  Number:	285070
  Issued:	5 December 1998
  Valid to:	5 December 1999
  Total airframe hours:	16,775.3 hrs

• Weight and balance
  The aircraft weight at the time of the occurrence was about 2,759 kg. The maximum allowable take-off weight was 2,994 kg. The centre of gravity was within limits.

• Maintenance history
  An examination of the maintenance history of XFF revealed that the aircraft had been inspected on 6 November 1998 by an airworthiness officer from the Civil Aviation Safety Authority. As a result of the inspection, Aircraft Survey Report (ASR)111642 was issued to the maintenance organisation. The report listed five Code B and one Code C defects. Persuant to Civil Aviation Regulation 38(1) the maintenance organisation was required to assess and rectify Code B defects as necessary. CASA form ASSP 604 states that "An endorsement of the maintenance release in accordance with Civil Aviation Regulation 50 may be required". Yes and no boxes, on the Aircraft Survey Report (ASR)111642 dated 6 November 1998, to indicate whether maintenance release endorsement was required, were not entered. Code C defects constitute "a contravention of requirements imposed under the Civil Aviation Regulations" and were required to be assessed and rectified as necessary.

  The defects were:

  • an oil leak in the left engine - Code B;
  • the left landing gear torque links were worn at the pivot points - Code B;
  • cracks in the left landing gear cowling - Code B;
  • surface corrosion on the underside of the left wingtip - Code B;
  • a broken bonding wire on the right flap - Code B; and
  • there was no load limitation placard on the rear baggage door - Code C.

  The aircraft logbook recorded that the last Schedule 5 (100 hourly) maintenance on the aircraft was completed on 5 December 1998. There was no record that the defects notified in ASR 111642 had been rectified during the maintenance. The engineering manager had certified an entry in the logbook regarding the inspection. It stated that there were no defects noted during the maintenance. The aircraft was flown the following day.

  An entry in the aircraft logbook dated 02 January 1999 listed the following maintenance actions:

  • An oil leak from the left engine was rectified by the removal of the engine sump, replacement of the sump gasket and re-fitment of the sump assembly.
  • Surface corrosion on the left-wing tip was repaired.
  • Stop drilling was conducted to control cracks in the left main landing gear leg fairing.
  • A left engine cowl latch was replaced.
  • Both left and right magnetos on the left engine were replaced with overhauled units. This was done for convenience as the replaced units were approaching the end of their in-service lives.
  • The left engine lower mounts were replaced.
  • The engine dual tachometer (RPM) instrument was repaired and refitted to the aircraft.
  • Support brackets were fitted to the left engine exhaust.
  • Bonding wire on the right-hand flap was replaced.
  • The right engine starter was lubricated.
  • Engine intake ducting to the left engine was replaced because of oil contamination.

  There was no record of any maintenance being conducted on the left landing gear torque links.

  At the time the maintenance was carried out, the aircraft had completed 19.6 flying hours since the issue of the maintenance release on 5 December 1998. The aircraft then completed a further 43.9 flying hours before the commencement of the accident flight. There was no record of any maintenance action being undertaken during this intervening period relating the rectification actions or any other matter.

Meteorological information

The Bureau of Meteorology advised that the probable weather at Coconut Island around the time of the accident was as follows:

• Isolated to scattered showers, and isolated thunderstorms;
• North-westerly wind at about 15 kts;
• Generally good visibility but reducing in precipitation; and
• Broken cumulus cloud with a base at 2,000 ft, with broken higher layers.

At 1400, the automatic weather station at Coconut Island recorded an ambient temperature of 30 degrees Celsius, a dew point of 25 degrees Celsius, and an atmospheric pressure of 1007 hectopascals. Witnesses at the island reported that the weather was fine at the time of the accident, with the wind gusting from the northwest.

Aids to navigation

Not relevant

Communications

The pilot was communicating on the area frequency of 120.3 MHz during the flight. The pilot of another aircraft heard transmissions from the pilot of the Islander on that frequency.

Aerodrome information

Coconut Island is about 110 km NE of Thursday Island. The island is composed of coral sand and is predominantly flat. It extends east-west for about 1.75 km, and is less than 0.5 km across, north-south, at its widest part. The airstrip occupies the eastern portion of the island and is aligned east-west. It is 880 m long, 60 m wide and composed of grassed coral sand. On the southern side of the strip, and extending for most of its length, is a sand dune approximately 5 m high with coconut palms growing on it.

The threshold for runway 27 is about 350 m from the eastern extremity of the island. At the time of the accident, the local refuse tip was situated between the end of the strip and the eastern extremity of the island, and north of the extended centreline of the runway. A dirt road linked the community living area and the refuse tip. The road followed the southern side of the strip to the eastern end before turning north towards the refuse tip area.

Flight recorders

The aircraft was not equipped with flight data or cockpit voice recorders, nor was such equipment required by regulation.

Wreckage examination

The wreckage was subjected to tidal saltwater immersion for 3 days before it was examined.

• Airframe
  An examination of the airframe did not reveal any fault that might have contributed to the accident. All flying controls were capable of normal operation prior to impact. The wing flaps were in the full-down position at impact. The right-wing fuel tank had been ruptured by the impact, while the left-wing tank was intact. A significant quantity of fuel remained in the left tank.

• Cabin
  The pilot's seat was mounted on a frame attached to the cabin floor. The seat could be adjusted fore and aft on the frame, but there was no vertical adjustment. During the impact, the frame partially collapsed down and towards the right. The seat was locked in the full-forward position. The pilot's lap-sash harness assembly remained intact during the impact.

  The rudder pedals were adjustable fore-aft into a locked position as selected by the pilot. The rudder pedals on the left side of the cockpit were locked one notch forward of the rearmost position. Damage indicated that the pedals were locked in that position at impact.

  The cabin was fitted with four bench-type passenger seats, each capable of seating two persons. The seat frames were secured to the floor. Two lap safety harnesses were attached to each seat frame.

  At the initial examination of the wreckage, there were no passenger seats in the cabin. All seats had been removed from the cabin during the rescue activities. One seat, found above the high-water mark, was recovered for examination. The remaining seats were not found and probably disappeared as a result of tidal action.

  Those involved in the initial response following the accident indicated that one seat remained attached to the cabin floor and was levered free with a crowbar. The remaining seats were loose, apparently after becoming detached during the impact sequence. Examination of the seat attachment points indicated that the first, second and third row seat frames had failed due to impact induced stresses. There was significant bending forward and to the right. Examination of the seat found above the high-water mark indicated that it was the rear seat that had been levered from the floor during the rescue activities.

  Apart from the two safety harnesses attached to the rear seat, only one-half of one other passenger harness was recovered. A section of a broken seat attachment bracket remained attached to the harness. The original location of the harness piece could not be determined.

• Engines and propellers
  Both engines and propellers were recovered from the accident site and examined. Disruption of the airframe prevented determination of the position of the engine controls at impact.

  The right propeller exhibited signs of severe tip curl and leading-edge abrasion, consistent with the engine developing high power at impact. Examination of the engine did not reveal any condition likely to have prevented normal engine operation. Salt-water corrosion damage prevented a detailed examination of the carburettor.

  The left propeller showed little evidence of rotational damage. The propeller had not been feathered. Laboratory examination of a failure of the left engine mixture control rod confirmed that the failure occurred at impact as a result of impact induced stresses. Examination of the engine did not reveal any condition likely to have prevented normal operation. After sand and other internal debris were removed, the magnetos were bench run for more than 30 minutes. They functioned normally during that period. The condition of the carburettor prevented confirmation of its serviceability at impact.

• Carburettor heat system
  The left and right carburettor heat control levers were mounted on the lower quadrant of the cockpit centre pedestal. Both levers had been bent flat against the pedestal face, and were in the OFF position.

  The carburettor air intake system of each engine had been destroyed during the impact sequence. Neither the pre-impact position of the normal/alternate air doors, nor the condition of the hot air flexible hose, could be determined.

Impact information

Consideration of the wing and nose section crush lines, along with the nature of damage to the fuselage and horizontal stabiliser, indicated that the aircraft was yawing and rolling left at impact. The pitch attitude at impact was 40-50 degrees nose-down. The right wing struck the ground first and bore the principal impact. The nose section, and then the left wing outboard leading edge struck the ground. Because of tidal activity, no ground impact marks were evident. The aircraft speed at impact could not be determined.

Medical and pathological information

The Bureau had not received the medical and pathological information at the time of the release of this report.

Fire

There was no fire.

Survival aspects

The deformation of the nose section and the forward/downward rotation of the wing significantly reduced the occupiable cockpit space. This, along with the impact forces, meant that the chances of survival for the pilot were low.

The surviving passenger indicated that she occupied the seat row immediately behind the pilot. The other two passengers occupied the second and third rows. The failure of the seat-to-floor attachments of the occupied seats in the aircraft cabin indicated that deceleration forces experienced in this area were high, thereby reducing survivability.

Aircraft operation

• Emergency operating procedures
  Section 4 of the Owner's Handbook for the aircraft type addresses emergency operating procedures. Relevant extracts from the section include the following:

  "Warning ...
  It is essential to raise the flaps to the fully up position to achieve the optimum climb gradient."

  "Critical engine
  Failure of the left engine has the most adverse effect on the handling and performance of the aircraft."

  "Landing with one engine inoperative
  Make an initial approach to approximately 65 kt (75 m.p.h.) IAS with the flaps selected to TAKE-OFF (25 deg). When committed for landing, select FLAPS DOWN (56 deg) and reduce speed over the threshold to a value compatible with the information scheduled in Sect. 6 and touchdown normally."

  Section 3, Operating Instructions, of the Owner's Handbook, included the following information:

  "Touch down
  Initial approach should be made at 65 kts (75 m.p.h.) IAS with flaps at TAKE-OFF (25 deg). After selection of FLAPS DOWN (56 deg) the speed may be progressively reduced to the appropriate threshold speed quoted in Section 6. After touch down allow the nose wheel to sink gently and apply the brakes as required."

  "Baulked landing
  Apply full power smoothly to the engines and be prepared to deal with a nose-up change in trim which can require a strong stick force, especially if the airspeed is low. Establish a positive climb away, select flaps to T.O., trim the aeroplane and accelerate to 61 kts (70 m.p.h.). Select flaps UP at a height above 200 feet and climb out at 65 kts (75 m.p.h.) IAS."

• Carburettor icing
  Section 3, Operating Instructions, of the Owner's Handbook, included the following information:

  "260 H.P. ISLANDER

  Use of carburettor heat
  Carburettor icing can occur, unexpectedly, in various combinations of atmospheric conditions. On damp, cloudy or foggy days, regardless of the outside temperature, keep a sharp watch for power loss, indicated by a decrease in manifold pressure. When this is seen, apply full carburettor heat for 30 seconds; this action will cause a further slight drop in manifold pressure. Return the heat control levers to OFF and note that selected engine power is restored. Do not keep heat selected FULL for long periods or excessive power loss will result, with very little indication from the manifold pressure indicator. During normal flight operations the carburettor heat control levers should be left in the OFF position."

  Section 3 also included, in the "Airfield Approach" checklist, the following comment on carburettor heat:

  "Intermittent use may be advisable to ensure responsive engines if a baulked landing is likely and ambient conditions are such that ice formation could occur."

  The temperature information supplied by the Bureau of Meteorology for Coconut Island around the time of the accident indicated that the atmospheric conditions were conducive to light carburettor ice forming at cruise or descent engine power settings.

  The company chief pilot knew of no instance of carburettor icing in Islander aircraft operating in the Torres Strait. The normal practice was that company pilots did not use carburettor heat during flights in the area. Similar comment was received from other organisations and pilots with extensive experience in operating Islander aircraft in the Torres Strait area.

• Aircraft wing flap operation
  The Owner's Handbook, Section 2, titled Design Information, under the sub-heading Flight Controls, contained the following information:

  "Electrically operated single-slotted flaps are fitted. An actuator on the wing rear spar operates the flaps through a system of push-pull rods. A selector switch on the pilot's console controls the actuator and a flap position indicator is situated on the cabin roof instrument panel. The flap control selector switch is a spring-loaded centre OFF unit and is wired to the actuator through a system of relays. Moving the switch to the DOWN position will only move the flaps 25 degrees to a TAKE-OFF setting and when this setting has been reached a second downward switch movement will be required to set the flaps to DOWN. Similarly, when raising the flaps, the first switch movement will only raise them to the TAKE-OFF setting and a second switch movement is necessary to completely raise the flaps."

  Pilots who had flown the aircraft indicated that the flap selector switch had to be held up or down against the spring, for a short time, before flap movement commenced.

Aircraft single-engine climb performance

Section 1 of the Owner's Handbook for the aircraft stated that the minimum control speed (single engine) was 39 kts. It applied when the flaps were up and the propeller on the inoperative engine was feathered.

Section 6 of the Owner's Handbook contained aircraft performance data, including single engine rate of climb data at 65 kts with the flaps up. The data indicated that, at an aircraft weight of 2,727 kg, an ambient temperature of 30 degrees Celsius, and at sea level, the rate of climb the aircraft was capable of achieving with one engine inoperative was about 160 ft/min.

The aircraft manufacturer advised that there were no actual performance figures available for the BN-2A-26 Islander aircraft with one engine inoperative, propeller unfeathered, and flaps down. However, there were unofficial climb figures for a BN-2B-26 variant of the Islander, with flaps up, and an unfeathered propeller. These were measured under test conditions at 65 kts airspeed and indicated that there was a decrement of between 70 and 90 ft/min (depending on the unfeathered propeller RPM) below the scheduled one-engine inoperative performance figures. The manufacturer also advised that aircraft performance with both engines operating was reduced by approximately 40 per cent when the flaps were selected from up to down, although this data could not be applied directly to flight with one engine inoperative. Go-around tests with one engine inoperative and flaps down had not been conducted.

Pilots experienced on the aircraft type, reported that the performance of the aircraft with flaps down and one propeller not feathered was unlikely to allow a successful go-around to be conducted.

Other information

• Information from surviving passenger
  Approximately 6 months after the accident, the surviving passenger provided the following information concerning the flight.
  1. She was seated in the row behind the pilot.
  2. One of the other passengers was in the second row, while the third passenger was in the third row.
  3. Her safety harness remained secured throughout the flight.
  4. There was no unusual event during the flight: the engines sounded normal.
  5. When the aircraft was on approach to Coconut Island, the pilot said that they could not land because there was a truck on the airstrip.
  6. The passenger saw a vehicle on the strip. It was stationary, and near the eastern end of the strip.
  7. The pilot was cross and said that there was no driver in the vehicle.

• Other information from witnesses at Coconut Island
  At the time of the accident, an aircraft operated by another company was parked at the western end of the airstrip. Two pilots were loading a consignment of crayfish onto the aircraft. Neither saw or heard XFF arrive in the circuit or fly the approach, nor could they recall if a vehicle had been on the strip around the time the aircraft was on approach. They indicated that their loading activities, along with the existing wind conditions, would have greatly reduced the likelihood of them hearing sounds from the eastern end of the airstrip. They were not aware of the accident until one of the island residents who witnessed the accident from the eastern end of the island raised the alarm. They proceeded to the accident site and, along with some of the island residents, provided assistance to the victims as far as they were able. One of the island residents advised the Thursday Island Police of the accident. They arranged for a medical team and police to be flown to the island in two helicopters. They arrived at the island between one and one and one-half hours after the accident.

  None of the three island residents who witnessed the accident reported seeing a vehicle on the airstrip when the aircraft was on final approach.

• Information from other company pilots
  Other pilots working for the operator indicated that island airstrips within the Torres Strait area were generally free of obstacles for their operations. There had been occasions, however, when vehicles, persons, or animals on the airstrip had caused pilots to go-around from a landing approach, requiring them to make a second approach.

  There were no radio links between aircraft and persons at the island airstrips. Local populations relied on hearing and/or seeing aircraft arriving to become aware of their presence. Depending on the weather and wind conditions, pilots did not always overfly airstrips before joining the circuit but often joined the downwind leg before completing a base leg and landing off final approach.

• Birdlife on the airstrip
  In the period during which the accident occurred, there were large numbers of migratory birds on Coconut Island. Many hundreds were seen occupying the grassed runway area. The birds were small and difficult to see in the ankle-high grass. When approached by a vehicle, they generally remained on the ground until the vehicle was closer than 20-30 m. When they did fly, it was as a flock.

  The opinion of company pilots was that the birds were not sufficiently large to constitute a significant safety hazard to aircraft operations. They believed that the pilot would not have discontinued the approach because of bird activity given their small size, and given that an aircraft would normally have been almost at the point of touchdown before the birds would begin to fly. However, there was no evidence that the aircraft had struck a bird.

A Fokker 50 was conducting a practice Instrument Landing System (ILS) approach for runway 24 at Perth, with an intended overshoot to Cunderin. A Cessna 172 was tracking via Northam to Perth with an intended overshoot to Jandakot. Other traffic in the area at the time was a helicopter tracking from Perth to Mundijong.

At 1133 Western Standard Time the crew of the Fokker was conducting the overshoot onto a radar heading which placed the aircraft in close proximity to the Cessna. The pilot of the Cessna sighted the Fokker and turned and descended to avoid that aircraft.

Perth Air Traffic Control had two runway specific traffic management plans. These were North-East (duty runways 03/06) and South-West (duty runways 21/24). Airspace ownership changed dependent on the plan in use and the airspace was divided as Terminal Control Area South (TMA S) and Terminal Control Area North (TMA N), the division occurring along the 281/077 radials from the Perth VOR. At the time of the occurrence the plan placed North-East and TMA S under the control of approach (APP) and TMA N under the control of departures (DEP).

The Perth Aerodrome Controller (ADC) had assumed responsibility for the Aerodrome control position approximately 10 minutes prior to the occurrence. The departure controller had received a handover 10 minutes prior to the incident and was unaware that the Fokker would turn back towards the Cessna.

The relative positions and intended tracks of both the Fokker and the Cessna were such that the departure controller was required to maintain vertical separation until within approximately 5 NM of the aerodrome, with the Fokker passing under the Cessna near Parkerville.

The aerodrome controller coordinated with the approach controller for departure instructions for the helicopter. The approach controller issued the instruction "right unrestricted you separate all the inbounds". The aerodrome controller accepted this and the instructions were issued without reference to the departure controller. The aerodrome controller subsequently coordinated with the approach controller for the overshoot instructions for the Fokker. The approach controller issued the instruction "left 120 unrestricted", but shortly after revised the instruction to, "separate with the helicopter, or keep him on runway heading for a bit to get him above". The aerodrome controller advised that the left turn onto a heading of 120 degrees would suffice. The approach controller issued the overshoot instruction without reference to the departure controller.

Coordination between the tower controllers and the Terminal Control Area controllers utilises the Operational Data Information contained within the radar label display. Coordination between approach and departures is via hotline communications.

Temporary Local Instruction (TLI) SDW/98/160 page 47 Section 5 paragraph 12.3 provides a choice of units to coordinate for aircraft overshooting Perth with the proviso "as appropriate." That section was contained in a letter of agreement between the Tower and Terminal Control Area and specifically dealt with overshooting aircraft.

The same TLI at page 65 Section 6 paragraph 2.1.1 advised that "In all instances the next call must be to DEP". This instruction was headed "Management of Departing IFR Aircraft-Perth" and dealt with Perth Departures from the non-duty runway. That paragraph specifically required the aerodrome controller to coordinate with the departures controller for departure instructions on aircraft departing from the non-duty runway but did not mention the procedure to be followed for overshooting aircraft from the non-duty runway.

The pilot was conducting a visual flight rules (VFR) flight from Walgett to an airstrip near Merriwa. The aircraft had departed from Walgett earlier in the day, but had returned a short time later when it was reported that the weather at the destination was not suitable for VFR flight. The pilot felt that he was under pressure to complete the flight that day. He continued to monitor the weather by telephoning for weather reports that were available from an automatic Bureau of Meteorology outlet, and by contacting a friend near the destination airfield. The aircraft later departed at about 1415. A search was subsequently initiated when the aircraft failed to arrive at its destination. The wreckage of the aircraft was located two days later on the top of a ridge, 3,880 ft above mean sea level (AMSL), slightly to the left of the direct track between Walgett and Merriwa.

The aircraft was found to have collided with trees during a right turn, at a rate of descent of about 2,500 ft/min. The impact severed the outboard section of the right wing. The aircraft had then collided with other trees before striking the ground. The right fuel tank had ruptured during descent through the trees and an intense post-impact fire had consumed the cabin area and the fuselage section immediately behind the cabin. Although the accident was survivable, both the pilot and passenger received extensive burns while escaping from the burning wreckage. The pilot died sometime later from his injuries, before the aircraft was located by search-and-rescue services personnel during the morning of the second day of the search. A fixed emergency locator transmitter (ELT), mounted in the aft cabin area of the aircraft, was destroyed by the fire. While it was not possible to determine if the ELT had activated during the accident sequence, no signal from the ELT had been received by the satellite monitoring system. The pilot was known to possess a personal ELT; however, this was not located after the accident.

Examination of the wreckage did not reveal any deficiencies that were likely to have contributed to the accident. Data extracted from a portable global positioning system unit found at the accident site confirmed that the aircraft had been in a right turn when it collided with the trees. Shortly after the accident the pilot had written a brief message on the left tailplane of the aircraft. That message indicated the pilot's perception of the accident sequence, and was generally consistent with the analysis by the investigation team.

The pilot held a private pilot licence for aeroplanes, and a commercial helicopter licence, together with valid medical certificates; however, he did not hold a rating for flight in instrument meteorological conditions (IMC), nor was the aircraft approved for flight in IMC.

Reports from National Park rangers who were in the area at the time of the accident, and from the Bureau of Meteorology, indicated that the cloud base was 3,600 ft AMSL, and that cloud was covering the ridge where the wreckage was found. The weather over lower terrain to the south-west of the accident site was reported to have been suitable for VFR flight.